Implementation of an Autonomous Patrol and Risk Detection Module based on WebRTC and ROS2 for a Quadruped Robot

This paper implemented a next-generation intelligent autonomous patrol system that can be integrated in complex and unstructured environments using the four-legged walking robot Unitree GO2.This is not an existing fragmentary module implementation, but an integrated system.First, WebRTC and ROS2 were integrated to enable low-latency video streaming and remote control within 200 ms.Second, ROS2-based SLAM and Nav2 were used to actively move along designated patrol routes and avoid obstacles to complete tasks.Third, an object detection model was installed to detect threats (fire, smoke, fallen people, etc.) in real-time during patrol, and the robot's location assistance and ArUco marker on the visitor's recognition table were recognized.Fourth, it is characterized by the construction of message services such as event occurrence and robot status inquiry using the remote control dashboard function and the Discord API to have the end-to-end functions required for security-safety scenarios.